Oracle introduced its Virtual Compute Appliance, an integrated, “wire once” stack for the data center that integrates compute, network, and storage resources in a software-defined fabric. It is designed for rapid deployment of both infrastructure hardware and application software, and runs runs Linux, Oracle Solaris, or Microsoft Windows.
The Oracle Virtual Compute Appliance can be scaled linearly, one server at a time, from 2 to 25 compute nodes
per rack. Oracle VM Templates enable application scalability across virtualized resources. Oracle Virtual
Compute Appliance controller orchestration software automatically powers up, installs, and configures the hardware and software environment. The appliance includes Oracle SDN software for virtualizing network resources.
Infiniband: Each Oracle Virtual Compute Appliance hardware configuration contains multiple redundant QDR InfiniBand switches and Oracle Fabric Interconnect systems that serve as gateways to the data
center’s Ethernet network.
x86: Compute nodes comprise Oracle’s Sun Server X3-2 systems containing Intel Xeon CPUs, high-speed dual inline memory modules (DIMM) memory, redundant, 40 Gb/sec InfiniBand host channel adapters (HCAs), and redundant disks
Storage: Oracle Virtual Compute Appliance features a fully integrated, enterprisegrade Oracle ZFS Storage Appliance for centrally storing the management environment as well as providing data storage for VMs.
“The Oracle Virtual Compute Appliance is unique in accelerating not only virtual infrastructure deployment, but also in speeding deployment of the complete application stacks through Oracle VM Templates,” said Wim Coekaerts, senior vice president, Linux and Virtualization Engineering, Oracle.
http://www.oracle.com/us/corporate/press/1988596
In April 2013, Oracle introduced its Virtual Networking open architecture for simplifying data centers using software-defined networking principles.
With Oracle Virtual Networking, the company is promising “the industry’s fastest data center fabric with up to 80 Gbs bandwidth to the server and support for Oracle’s SPARC T5, T4 and M5 servers along with the Oracle Solaris 11 on both SPARC and x86 platforms.
More specifically, Oracle Virtual Networking is a data center fabric for controlling larger server pools with scalable I/O. By using this architecture and its own SPARC T5 systems, Oracle can scale up to 1,000 servers and 128,000 cores of compute performance. The virtualization capabilities provide the ability to run more VMs per server and get predictable performance for applications with granular quality of service controls.
For cloud deployments, Oracle said its solution can create up to 16,000 private Ethernet layer-2 networks inside a single fabric and maintain necessary network security in multi-tenant cloud environment. Additionally, the new release of Oracle Fabric Manager 4.1 unifies management for SPARC and x86 systems deployed with the Oracle Fabric Interconnect.
Products in the Oracle Virtual Networking family include:
- Oracle Fabric Interconnect (formerly the Xsigo Fabric Director)– a switch offering 20 Quad Small Form-factor Pluggable (QSFP) connectors for connection to servers, storage, or Sun Data Center InfiniBand Switch 36 systems.
- Oracle Fabric Manager
- Oracle Fabric Monitor
- Oracle SDN
In July 2012, Oracle acquired Xsigo Systems, a start-up in San Jose, California that developed network virtualization technology. The Xsigo Server Fabric aims to do for infrastructure what VMware did for the servers — namely, to enable one-click network connections from virtual machines to any data center resource – including servers, networks, storage, and other virtual machines.
The Xsigo Server Fabric is a rack-based solution that works by virtualizing connections between networks, servers and storage, not by re-configuring switches, switch ports, or VLANs. It supports Ethernet and Fibre Channel connections at up to 40 Gbps. By providing this “one click” tool to connect VMs to networks and storage while using existing Ethernet routers + FC directors, Xsigo calculates that it can reduce the number of I/O cables and cards in a data center by up to 70%, thereby significantly cutting deployment CAPEX and simplifying operations.
